Rm. Kay et al., INTEROBSERVER AND INTRAOBSERVER VARIATION IN STAGING PATIENTS WITH PROVEN AVASCULAR NECROSIS OF THE HIP, Clinical orthopaedics and related research, (307), 1994, pp. 124-129
Diagnosis and treatment of avascular necrosis of the hip has long been
predicated on the evaluation of plain radiographs, though other modal
ities (such as magnetic resonance imaging) are being increasingly used
to aid in this evaluation, The Ficat classification is commonly used
to assess plain radiographs and to help determine what treatment is ap
propriate for a given patient. It also is used to help evaluate patien
t outcome after surgical or nonsurgical treatment. This study was desi
gned to evaluate the adequacy of plain radiographs in the evaluation o
f avascular necrosis of the hip. The plain radiographs of 25 hips with
avascular necrosis were analyzed on 3 occasions by 6 readers (450 tot
al readings). A clinically significant difference in radiographic stag
ing was defined as a Stage I or II reading of the radiographs of a giv
en hip on 1 reading and a separate reading of the same radiographs as
Stage III or IV. By this definition, interobserver variability for the
first reading resulted in clinically significant differences in 10 (4
0%) of the 25 hips. Intraobserver variability resulted in clinically s
ignificant differences for 10 (40%) of the 25 hips as well. The most e
xperienced readers in the study (a total joint specialist and a muscul
oskeletal radiologist) were consistent internally in their readings in
90% of cases, and had a clinically significant difference in only 1 c
ase (1%); however, these same 2 readers disagreed with one another on
the staging of 9 (36%) of 25 hips, with 4 (16%) of 25 hips having a cl
inically significant difference in readings. The kappa statistic for i
ntraobserver and interobserver variability was 0.82 +/- 0.16 and 0.56
+/- 0.01, respectively. This high degree of variability suggests that
plain radiographs alone are often inadequate for evaluating avascular
necrosis of the hip.